Research Of Highway Traffic Incident Detection Based On Evidential Reasoning Algorithm

In recent years, with the development of highway construction, it took great convenience to people’s life and travel, at the same time security issues occured on the highway also increasingly attracted people’s attention. All kinds of traffic incidents happened on the highway, would bring serious challenges to people’s life and property safety. Therefore, this made people to concern about rapid detection, timely processing of traffic incidents to reduce the harm of traffic incidents. And of course, as a core research, the traffic incident detection algorithm was attracted by many experts and scholars, and had a very rich algorithm results.Based on the above background, this paper relies on the national natural science foundation of "Accident risk assessment and safety design parameters optimization of highway traffic flow under the influence of visibility (No.51578207)", the ER algorithm is introduced to the traffic incident characteristics analysis and design process of incident detection algorithm.This paper firstly studies multi traffic incidents that occurred under the condition of different number of lanes, weather, visibility and temperature, to explore the impact when traffic incidents happened under different conditions. The P value is introduced to characterize the traffic incident characteristic, and explore the general rules of traffic incident characteristics under different conditions. And based on this ER algorithm is used for fusion evaluation of traffic incidents characteristics under various conditions, distribute the trust relationship of significant degree under various conditions.Secondly, on the basis of the above research, according to the relationship of the traffic incident characteristics after the ER algorithm fusion, the paper proposed the factors of the traffic incident characteristics under different conditions, then as a parameter it was introduced into traffic incident detection algorithm to improve the effect of traffic incident detection. Based on this the paper designed three kinds of traffic incident detection algorithm:multi-parameters identification algorithm under various conditions, incident detection algorithm based on Bayesian network under various conditions and incident detection algorithm based on SVM under various conditions.And based on this, to make the incident detection better and improve the applicable conditions of algorithm, the detection results of three basic algorithms were fused by ER algorithm, and then the paper established the traffic incident detection algorithm based on ER algorithm.Finally, with real traffic data, incident data, and the condition data, the paper studied different sets of traffic incident detection under various conditions, analyzed operation index of each algorithm, evaluated the effect of different algorithms, and at last drew a conclusion of this paper.This paper introduces ER algorithm in the study of traffic incident detection, introduces multi attribute decision-making method of ER algorithm to the research of traffic incident detection, and verifies that the traffic incident detection algorithm based on ER algorithm has good detection results with the real data. It provides reference of incident detection for future research.